Research

Politicians, pundits, and ordinary citizens frequently argue that private money has an outsized influence on U.S. politics. Yet research on the effect of money on politicians' behavior finds limited effects, especially on voting behavior. I construct a formal model to show that these limited effects may be an artifact of the institutional powers of agenda setters in Congress, and the strategic nature of the data generating process. The model predicts that under general conditions, vote buying against proposals can occur with no vote held, and that vote buying is most likely when uncertainty is high, the interest group and agenda setter are extreme, and the pivotal legislator is moderate. I argue that these conditions held during the 2021 negotiations over the Build Back Better Act. Using a synthetic control design, I test whether the swing voters in the Senate received more campaign contributions than they would have otherwise, finding evidence consistent with the model.

Spatial models of agenda setting are central to the study of legislative politics, yet they typically abstract from the influence of interest groups. To address this gap, I unify the classical one-dimensional agenda setting framework with models of vote buying, starting with one group and extending to two moving sequentially. Anticipating the groups' responses, the agenda setter strategically adjusts proposals to take advantage of support or to preempt opposition. This unification generates a number of insights into interest groups' behavior, and how their influence affects gridlock, coalition size, and policy moderation. Most notably, the model offers a new explanation for supermajority coalitions: rather than arising from competition between vote buyers, supermajorities emerge because the proposer biases policy toward the opposing group to reduce the likelihood of opposition. Contrary to previous theories, coalitions can be larger when there is just a single interest group than when there are two. 

Ansolabehere, de Figueiredo, and Snyder (2003) famously argue that most money in politics is motivated by consumption rather than investment, citing evidence that campaign contributions have little effect on legislators' voting behavior. We revisit this influential claim using updated data and new theoretical tools. First, we show that a comparable analysis using more recent data reveals a modest but statistically significant relationship between contributions and vote choice. Second, we use simulated data from a formal model of vote buying with an endogenous agenda to demonstrate that standard empirical approaches may systematically underestimate the effects of contributions. Third, we show that the challenge of inference becomes even more difficult when contributions are driven by a mix of expressive and instrumental motives. Finally, we review theoretical arguments that help explain why contributions might appear low even when they reflect attempts to influence policy. Together, these findings suggest that the influence of organized interests may be more substantial, and more difficult to detect, than early analyses implied.

Billions of dollars are spent lobbying the federal government each year, yet the effects of regulating this activity remain poorly understood. I develop a formal model to examine one such intervention: a cap on lobbying expenditures. In the model, expenditures serve purely as costly signals, or "burning money," used to convey a lobbyist’s intensity of preference. Lobbyists represent constituencies with known ideal points but hold private information about their sensitivity to policy deviations. A policymaker, uncertain about these types, chooses a policy that maximizes the sum of the lobbyists’ utilities. I show that in the absence of regulation, lobbyists can fully reveal their type in equilibrium. Surprisingly, imposing an exogenous expenditure cap increases the lobbyists’ ex-ante expected utility by reducing the informativeness of low spending. These findings suggest that lobbying regulations can improve outcomes for lobbyists on average, even when they restrict forms of influence.

It is commonly argued that if a market for political favors exists, then returns should be highest when elections are close because candidates have an increased willingness to sell favors. This simple theoretical argument justifies the burgeoning empirical literature using close-election regression discontinuity designs (RDDs) to test whether firms earn returns to campaign contributions. I develop a formal model to show  that there is a flaw in the standard theoretical argument: it ignores the extensive margin. That is, more favors being sold induces more donors to become active, thereby diluting average returns. In equilibrium, average returns to the winner’s contributors can actually be lowest when elections are close. Thus, the small or null local average treatment effects identified by the close-election RDD studies may be highly misleading.